This study explore the possibility of structural breaks in daily realized volatility series of agricultural commodity futures

and conduct an out-of-sample forecast to explore the effects of structural break on the performance of ARFIMAX-FIGARCH models for the realized volatility forecast

concentrating on procedures that utilize a variety of estimation window sizes designed to accommodate the potential structural breaks. The results indicate that the realized volatility of agricultural commodity futures exhibits the properties of structural breaks

asymmetry

and double long memory. In addition

combination forecasts with time varying weights across individual forecast models estimated with different estimation windows performs well

and the nonlinear combination forecasts with weights chosen based on a nonparametric kernel regression and the linear combination forecasts with weights chosen based on non-negative restricted least squares and Schwarz Information Criterion appears to be the most effective methods for forecasting the realized volatility of agricultural commodity futures under structural breaks.